Development of an Automatic Musical Keys Notes Detection Technique Using Mir Toolbox

Musical key recognition has traditionally been a difficult endeavor for humans, especially for people who lack adequate musical training. An expert musician must put up a lot of time and effort to listen to music and translate it into music sheets, which may be impossible for a novice to attempt. However, automatic musical key recognition streamlines the transcription process and is crucial for Music Information Retrieval (MIR). Although, research on an automatic system for detecting musical keys is still in its early stages, the findings thus far have shown to be extremely instructive for both machine learning and music composition.This research embraced the use of the MIR toolbox, a toolbox created specifically to extract musical features from audio samples. The musical key detection problem was transformed into a form of comparison of the audio waveforms and other musical features of the audio sample in this research, as a significant approach to solve the problem of music transcription. The MIR toolbox database which has recordings and Musical Instrument Digital Interface (MIDI) files of isolated piano and guitar sounds, was used to collect the training and testing data.The model's accuracy level was assessed using a confusion matrix as a performance indicator, and it turned out that the model has a 68% precision and 69% sensitivity, which shows a relatively high accuracy. Therefore, this study provides a potential solution to detecting musical keys.